Calibration Test Sites Selection and Characterization Methods for Calibration/Validation and Test site requirements
Customer Contract No WP No
: ESA/ESRIN : 21125/07/1-OL : 222
Title
:
Abstract
:
Document Ref : Issue Date : Issue :
CALIB-TN-WP222-GAEL_ETH Issue 1.0
CALIB-TN-WP222-GAEL_ETH 16 March 2009 1.0
Calibration Test Sites Selection and Characterization Methods for Calibration/Validation and Test site requirements
GAEL-ETH contribution to WP 222 The objective of WP222 is to inventory the calibration validation methodologies in order to propose some requirements regarding the test field configuration. The test field configuration and equipment have been detailed in deliverable document of workpackage 210
Author
Distribution
:
Sebastien Saunier, Sultan Kocaman
Approval
: B. Berthelot
Accepted
: Ph. Goryl
: Hard Copy File: Filename:
CALIB-TN-WP222-GAEL_ETH_001.doc
Copyright © 2009 VEGA Group PLC All rights reserved. No part of this work may be disclosed to any third party translated reproduced copied or disseminated in any form or by any means except as defined in the contract or with the written permission of VEGA Group PLC. GAEL ETH Cité Descartes Tel: +33 (0)1 64 73 99 55 Fax: +33 01 64 73 51 60 www.gael.fr
VEGA Technologies SAS © 2008 VEGA Group PLC
Page 1 of 54
CALIB-TN-WP222-GAEL_ETH Issue 1.0
Calibration Test Sites Selection and Characterization Methods for Calibration/Validation and Test site requirements
This Page Is Intentionally Blank
Page 2 of 54
VEGA Technologies SAS
Calibration Test Sites Selection and Characterization Methods for Calibration/Validation and Test site requirements
CALIB-TN-WP222-GAEL_ETH Issue 1.0
TABLE OF CONTENTS 1. APPLICABLE AND REFERENCE DOCUMENTS .................................................................12 1.1
Reference Documents ........................................................................................................... 12
2. INTRODUCTION .....................................................................................................................16 3. DEFINITION.............................................................................................................................18 3.1
Fundamental concepts .......................................................................................................... 18
3.1.1
Calibration vs. Validation ............................................................................................... 18
3.1.2
Georeferencing approaches.......................................................................................... 18
3.1.3
Rigorous vs. generic sensor models ............................................................................. 20
3.1.4
Rigorous sensor model parameters .............................................................................. 22
3.2
Sensors’ geometric characterizations ................................................................................... 23
3.3
List of the Cal/Val parameters ............................................................................................... 26
3.3.1
Interior orientation parameters ...................................................................................... 27
3.3.1.1 Optical system parameters - Principal point and distance ....................................... 27 3.3.1.2 Optical system parameters – Lens Distortion Parameters....................................... 27 3.3.1.3 Imaging sensor parameters...................................................................................... 27 3.3.2
External orientation parameters .................................................................................... 27
3.3.2.1 Positional accuracy and stability .............................................................................. 27 3.3.2.2 Attitude measurement accuracy, and stability in time .............................................. 27 3.3.2.2.1
Inertial Measurement Systems (INSs) ............................................................ 27
3.3.2.2.1.1
Description ............................................................................................... 27
3.3.2.2.1.2
Errors........................................................................................................ 27
3.3.2.2.2
Star tracker ..................................................................................................... 27
3.3.2.3 Alignment.................................................................................................................. 27 3.3.3
Overall geometric accuracy parameters ....................................................................... 27
3.3.3.1 Direct Geolocation Accuracy .................................................................................... 27 3.3.3.2 Image-to-Image registration accuracy...................................................................... 27 3.3.3.3 Local coherence detection........................................................................................ 27 3.3.3.4 Stereoscopic measurement...................................................................................... 27 VEGA Technologies SAS © 2008 VEGA Group PLC
Page 3 of 54
CALIB-TN-WP222-GAEL_ETH Issue 1.0
Calibration Test Sites Selection and Characterization Methods for Calibration/Validation and Test site requirements
3.3.3.5 DSM product testing ................................................................................................. 27 3.3.3.6 Ortho product testing ................................................................................................ 27 3.3.3.7 Value-added product testing..................................................................................... 27 4. CALIBRATION/VALIDATION METHODS ..............................................................................27 4.1
Estimate of the interior orientation parameters ..................................................................... 27
4.1.1
Lens distortions (for optical sensors)............................................................................. 27
4.1.2
Imaging sensor parameters........................................................................................... 27
4.2
Estimate of the exterior orientation parameters .................................................................... 27
4.2.1
Attitude measurement accuracy, and stability in time ................................................... 27
4.2.2
Positional accuracy and stability ................................................................................... 27
4.2.3
Sensors’ relative alignments ......................................................................................... 27
4.3
Estimate of the overall accuracy............................................................................................ 27
4.3.1
Direct Geolocation Accuracy ......................................................................................... 27
4.3.2
Image-to-Image registration .......................................................................................... 27
4.3.3
Local coherence ............................................................................................................ 27
4.3.4
Stereoscopic measurement........................................................................................... 27
4.3.5
DSM Product testing...................................................................................................... 27
4.3.6
Ortho product testing ..................................................................................................... 27
4.4
Test field requirements according to the cal/val activity and the selected method ............... 27
4.4.1
Calibration ..................................................................................................................... 27
4.4.2
Validation....................................................................................................................... 27
4.5
Summary ............................................................................................................................... 27
APPENDIX 1: LIST OF THE EXISTING PROGRAM ....................................................................27 APPENDIX 2: THE GEOMETRIC CALIBRATION PROCESS .....................................................27
Page 4 of 54
VEGA Technologies SAS
Calibration Test Sites Selection and Characterization Methods for Calibration/Validation and Test site requirements
CALIB-TN-WP222-GAEL_ETH Issue 1.0
List of tables Table 1: Table of the Reference Documents. ................................................................................. 12 Table 2: Sensor class ...................................................................................................................... 16 Table 3: Categorization of the satellite optical sensor geometries.................................................. 21 Table 4: Categorization of the satellite optical sensor geometries.................................................. 24 Table 5: Line geometries of the satellite Linear Array CCD sensors .............................................. 24 Table 6: List of parameters to be estimated.................................................................................... 27 Table 7: Error sources in GPS positioning (RD23) ......................................................................... 27 Table 8: Examples for the minimum number GCPs to be used in a rigorous model with trajectory modelling and self-calibration* ........................................................................................................ 27 Table 9: The number of unknowns in the widely used generic models and the corresponding minimum number of GCPs required for the solution (RD24). ......................................................... 27 Table 10: List of the parameters...................................................................................................... 27 Table 11: Calibration program......................................................................................................... 27
VEGA Technologies SAS © 2008 VEGA Group PLC
Page 5 of 54
CALIB-TN-WP222-GAEL_ETH Issue 1.0
Calibration Test Sites Selection and Characterization Methods for Calibration/Validation and Test site requirements
List of figures Figure 1: Transformation from camera system to the object (ground) system. PC: projection centre, PP: principal point, c: camera principal distance (focal length), (x,y,z): axes of image coordinate system, (X,Y,Z): axes of object space coordinate system .............................................................. 23 Figure 2: Idealized geometric image formation of an optical system. N, N′ = Nodal points = projection centres O, O′ (RD15)...................................................................................................... 27 Figure 3: Focal plane arrangement of the IKONOS sensor (RD36) ............................................... 27 Figure 4: Quickbird and SPOT HRG sensors, focal plane arrangement (RD36)............................ 27 Figure 5: Inertial measurement units (RD22). ................................................................................. 27 Figure 6: IKONOS stereo image collection (RD34) with asynchronous along-track stereo collection principle. .......................................................................................................................................... 27 Figure 7: Scan angles and the shape of the imaged geographical area for EROS-A1 VHR satellite sensor (RD35]). The images are acquired with asynchronous along-track stereo collection principle ......................................................................................................................................................... 27 Figure 8: Synchronous along-track stereo collection with the ALOS/PRISM sensor (www.jaxa.jp)27 Figure 9: SPOT-5 HRG viewing capacity both along-track and across-track (RD35]). .................. 27 Figure 10: ALOS/AVNIR-2 Interior observation concept (ALOS data users handbook)................. 27 Figure 11: Linear Array CCD chips are used for image acquisition of the IRS 1C/1D sensor (India). Each chip includes 4096 CCD pixels. About 96 pixels remain in the overlapping area of the chips. ......................................................................................................................................................... 27 Figure 12: Object space residuals (3D vectors) of GCP coordinates back-projected into the image space in Saitama testfield. The image is acquired with the ALOS/PRISM forward camera. The systematic behaviour in the residuals is caused because of missing calibration parameters (positions of the CCD detectors). The errors were corrected in using the self calibration procedure, [RD5]. .............................................................................................................................................. 27 Figure 13: Object space residuals in the Zurich/Winterthur testfield obtained from the Direct georeferencing with the given RPCs (RD41). Local systematic errors are observed in the residual distribution. ...................................................................................................................................... 27 Figure 14: Suboptimal GCP arrangement, most of the GCPs are located in close groups on the ground (Saitama testfield, Japan). .................................................................................................. 27 Figure 15: Overview of the GCP distribution in the PRISM nadir image of Sakurajima. The red circles represent the GCPs. The distribution is suboptimal due to partial coverage in planimetric extent of the image (RD41). ............................................................................................................ 27 Figure 16: Point distribution in the Zurich/Winterthur testfield image acquired with the ALOS/PRISM sensor. The red circles represent the GCP locations. The distribution is quite good for the analysis of IO and EO parameters, as well as overall accuracy.......................................... 27 Figure 17: ALOS platform configuration (ref: www.jaxa.jp)............................................................. 27
Page 6 of 54
VEGA Technologies SAS
Calibration Test Sites Selection and Characterization Methods for Calibration/Validation and Test site requirements
CALIB-TN-WP222-GAEL_ETH Issue 1.0
Figure 18: Details visible is more important than the pixel size to assess the image quality (RD33) ......................................................................................................................................................... 27 Figure 19: Key factors effecting the resolution and quality of orthoimages RD33. ......................... 27
VEGA Technologies SAS © 2008 VEGA Group PLC
Page 7 of 54
CALIB-TN-WP222-GAEL_ETH Issue 1.0
Calibration Test Sites Selection and Characterization Methods for Calibration/Validation and Test site requirements
AMENDMENT POLICY This document shall be amended by releasing a new edition of the document in its entirety. The Amendment Record Sheet below records the history and issue status of this document. AMENDMENT RECORD SHEET ISSUE
DATE
DCI No
Ed0.1
12/10/2008
N/A
Initial Issue
Ed0.2
21/11/2008
N/A
Internal review with Sultan K, second issue
Ed1.0
16/03/2009
N/A
Final version
Page 8 of 54
REASON
VEGA Technologies SAS
Calibration Test Sites Selection and Characterization Methods for Calibration/Validation and Test site requirements
CALIB-TN-WP222-GAEL_ETH Issue 1.0
This Page Is Intentionally Blank
VEGA Technologies SAS © 2008 VEGA Group PLC
Page 9 of 54
CALIB-TN-WP222-GAEL_ETH Issue 1.0
1.
Calibration Test Sites Selection and Characterization Methods for Calibration/Validation and Test site requirements
APPLICABLE AND REFERENCE DOCUMENTS The following is a list of documents with a direct bearing on the content of this report. Where referenced or applicable in the text, these are identified as RD.n or AD.n, where 'n' is the number in the list below
1.1
Reference Documents Table 1: Table of the Reference Documents. ID document
Document reference
[RD0]
Calibration Test Sites Selection and Characterisation. Technical proposal. VEGASPA.PRO.070197
[RD1]
Tadono T, Shimada M, Watanabe M, Hashimoto T, Iwata T, Calibration and Validation of PRISM on board ALOS, 2003.
[RD2]
G Dial ; Jaceck Grodecki, Test Ranges for Metric Calibration and Validation of Satellite Imaging Systems
[RD3]
Gachet, R. 2004, Spot5 In-Flight Commissioning: Inner Orientation Of HRG And HRS Instruments, ISPRS Congress Istanbul 2004.
[RD4]
Tadono T., Shimada M., Iwata T., TAKAKU J., 2007. Accuracy assessment of geolocation determination for PRISM and AVNIR-2 onboard ALOS. Proceedings of the 8th Conference on “Optical 3D Measurement Techniques”, Zurich, Switzerland, 9-12 July, Vol. I, pp. 214-222.
[RD5]
Gruen A., Kocaman S., Wolff K., 2007. Calibration and Validation of Early ALOS/PRISM Images. The Journal of the Japan Society of Photogrammetry and Remote Sensing, Vol 46, No. 1, pp. 24-38.
[RD6]
Lee D.S., Storey J.C., Choate M.J., 2004, Four Years of Landsat-7 On Orbit geometric calibration and performance, IEEE Transactions on Geoscience and Remote Sensing; Vol 42, Number 12, pp 2786-2795
[RD7]
B.Smiley, JACIE 2008 – The absolute and relative geolocation accuracy of Quick Bird and World View
[RD8]
D.Helder, K.Row, JACIE2006 - NASA / SDSU Geopositional Characterization
[RD9]
Saunier S., Goryl P. et al., 2007, The contribution of the European Space Agency to the ALOS PRISM / AVNIR-2 commissioning phase, IGARSS 2007.
[RD10]
New Third Party Mission, Quality Assessment Kompsat-2 Mission, Sébastien Saunier, Bernard Collet, Aboubakar Mambimba, GAEL-P232-DOC-005, May 2008.
[RD11]
Saunier. S, Goryl P., MERIS Full Resolution products, Geometric aspects, Proceedings of Meris AATSR Validation team Meeting, MAVT 2006. http://www.gael.fr/telimago/
[RD12]
Saunier.S, Goryl P., Meris Absolute Geolocation status 2002-2005 period, PCS 2006, http://earth.esa.int/pcs/envisat/meris/documentation/
[RD13]
Iwasaki A, Fujisada H, ASTER geometric performance, Transactions on Geoscience and Remote Sensing; Vol 43, Number 12
Page 10 of 54
VEGA Technologies SAS
Calibration Test Sites Selection and Characterization Methods for Calibration/Validation and Test site requirements
CALIB-TN-WP222-GAEL_ETH Issue 1.0
ID document
Document reference
[RD14]
MODIS Level 1A Earth Location: Algorithm Theoretical Basis Document Version 3.0; SDST-092 August 26,1997
[RD15]
Kraus, Karl., 1993. Photogrammetry Vol. I., Fundamentals and Standard Processes, Der. Dümmler Verlag, Bonn, Germany, ISBN 3-427-78684-6, 397 pages.
[RD16]
Atkinson, K.B. (Editor), 1996. Close range photogrammetry and machine vision. Whittles Publishing, Caithness, Scotland, 371 pages.
[RD17]
Clarke, T.A., Fryer, J.G. and Wang, X., 1998. The principal point and CCD cameras. The Photogrammetric Record, Vol. 16(92), pp. 293-312.
[RD18]
Brown, D.C., 1976. The bundle adjustment – progress and prospects. International Archives of Photogrammetry, Vol. 21(3), paper no 3-03-041, 33 pages.
[RD19]
Mikhail, E.M., Bethel, J.S., McGlone, J. C., 2001. Introduction to modern photogrammetry. John Wiley and Sons Inc., New York, U.S.A., 479 pages.
[RD20]
Poli, D., 2005. Modelling of spaceborne linear array sensors. Doctoral Thesis, No. 85, Institute of Geodesy and Photogrammetry, ETH Zurich, Zurich, Switzerland.
[RD21]
Nieke, J., Itten, K.I., 2007. Hyperspectral imagers. 12th SSOM Engelberg Lectures on Optics, 5–7 March, Hotel Regina Titlis, Engelberg, Switzerland (accessed online on http://www-optics.unine.ch/events/engelberg2007/ on 20.01.2008).
[RD22]
Grewal M. S., Weill L. R., Andrews A. P., “Global Positioning Systems, Inertial Navigation, and Integration”, John Wiley and Sons Publication, New York, U.S.A., 2001.
[RD23]
Jekeli, C., “Inertial Navigation Systems With Geodetic Applications: inertial Navigation Systems”, Published by Walter de Gruyter, ISBN 3110159031, 2000.
[RD24]
Gruen, A., Kocaman, S., 2008. Optical Sensors High Resolution: Geometry Validation Methodology. Technical report submitted to ESA/ESRIN, February, 219 pages.
[RD25]
Kocaman, S., 2008. Sensor Modeling and Validation for Linear Array Aerial and Satellite Imagery. Ph.D. Thesis, submitted to ETH Zurich, Institute of Geodesy and Photogrammetry, Zurich, Switzerland.
[RD26]
Seeber, G., 2003. Satellite Geodesy: Foundations, Methods, and Applications. Published by Walter de Gruyter, ISBN 3110175495, 9783110175493, 589 pages.
[RD27]
Kocaman, S., 2003. GPS and INS Integration with Kalman Filtering for Direct Georeferencing of Airborne Imagery, ETH. Zurich, Institute of Geodesy and Photogrammetry, 30 January.
[RD28]
Kocaman, S., 2008. Sensor Modeling and Validation for Linear Array Aerial and Satellite Imagery. Ph.D. Thesis, submitted to ETH Zurich, Institute of Geodesy and Photogrammetry, Zurich, Switzerland.
[RD29]
Wang, Z., 1990. Principles of Photogrammetry. Press of Wuhan Technical University of Surveying and Mapping, Beijing, China.
[RD30]
Gruen, A., Beyer, H.A., 2001. System Calibration Through Self-Calibration. Calibration and Sensor Orientation of Cameras in Computer Vision, Eds. Gruen, Huang, Springer-Verlag Berlin, Heidelberg, pp.163-193.
VEGA Technologies SAS © 2008 VEGA Group PLC
Page 11 of 54
CALIB-TN-WP222-GAEL_ETH Issue 1.0
Calibration Test Sites Selection and Characterization Methods for Calibration/Validation and Test site requirements
ID document
Document reference
[RD31]
Valorge, C., Meygret, A., Lebégue, L., Henry, P., Bouillon, A., Gachet, R., Breton, E., Léger, D., Viallefont, F., 2004. Forty Years of Experience with SPOT In-flight Calibration. In Post-Launch Calibration of Satellite Sensors – Morain and Budge (eds.), Taylor & Francis Group, London, England, pp. 119-133.
[RD32]
Cook, M.K., Peterson, B.A., Dial, G., Gerlach, F., Hutchins, K., Kudola, R., Bowen, H.S., 2001. IKONOS Technical Performance Assessment. White Paper, www.geoeye.com (last accessed on 09.05.2007).
[RD33]
Poon, J., Smith, L., Fraser, C., 2006. Orthoimage Resolution and Quality Standards. Project 2.1 Report submitted to Cooperative Research Centre for Spatial Information, Melbourne, Australia. http://spatialinfocrc.org/uploads/publications/PUBLICATION_306.pdf (last accessed on 24.10.2008)
[RD34]
Grodecki, J. Dial, G., 2001. IKONOS geometric accuracy. Proceedings of ISPRS Joint Workshop “High Resolution Mapping from Space” 2001, 19-21 September, Hanover, Germany, (CD-ROM).
[RD35]
Gleyzes, J.P., Meygret, A., Fratter, C., Panem, C., Baillarin, S., Valorge, C., 2003. SPOT5: System Overview and Image Ground Segment. Proceedings 2003 IEEE International Geoscience and Remote Sensing Symposium, Toulouse, France, 21–25 July vol. I, pp. 300–302.
[RD36]
Jacobsen, K., Baltsavias, E., Holland, D., 2008. Information extraction from high resolution optical satellite sensors. Tutorial-10, XXIst ISPRS Congress, Beijing, China, , July 3.
[RD37]
Cheng, P., Toutin, T., Zhang, Y., Wood, M., 2003. QuickBird — Geometric correction, path and block processing and data fusion. Earth Observation Magazine's (EOM), May, pp. 1–10.
[RD38]
Jacobsen, K., 2007. Comparison of Image Orientation by IKONOS, QuickBird and OrbView-3. EARSeL. Band "New Developments and Challenges in Remote Sensing". Rotterdam : Millpress, Warschau, pp. 667-676
[RD39]
Kocaman, S., Gruen, A., 2008. Geometric modeling and validation of ALOS/PRISM imagery and products. 21st ISPRS Congress, Beijing, China, 3-11 July, The International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences, Vol. XXXVII. Part B1, pp. 731-738.
[RD40]
P. V. Radhadevi, S. S. Solanki, 2008. In-Flight geometric calibration of different cameras of IRS P6 using a physical sensor model. The Photogrammetric Record 23(121): 69–89.
[RD41]
S. Leprince, 2008. In-Flight CCD Distortion Calibration for Pushbroom Satellites Based on Subpixel Correlation. IEEE Transaction on Geoscience and Remote Sensing, vol 46, No 9, Septembre 2009.
[RD44]
Mathias Schneider, Manfred Lehner, Rupert Müller, Peter Reinartz Stereo Evaluation of ALOS/PRISM Data on ESA-AO Test Sites - First DLR Results, ISPSRS, Beijing,2008.
[RD45]
D. Greslou, F. Delussy, Geometric Calibration of Pleiades Location Model, ISPSRS, MLV, 2006.
Page 12 of 54
VEGA Technologies SAS
Calibration Test Sites Selection and Characterization Methods for Calibration/Validation and Test site requirements
2.
CALIB-TN-WP222-GAEL_ETH Issue 1.0
INTRODUCTION This document has been designed according to RD0 and is the third deliverable in the frame of this project. Previously in the calibration site project, the on going geometric calibration/validation campaigns have been inventoried (Work package WP110). The next work package (WP210) has focused on the materials required to performed calibration / validation activities according to the sensor categories (Table 2). The scope of this document is the calibration/validation methods. The purpose of this study is two folds;
To output test site requirements on test site from the method itself,
To output test site requirements on test site from the calibration/validation methodology, these requirements are more specific to the sensor characteristics.
The document is starting in bringing details on fundamental concept of geometric calibration and on the list of the calibration validation parameter. The explanations on the methods follow and some first requirements on site configuration will be proposed. The sensor class is recalled with the table just here after. Table 2: Sensor class GSD Class
Error
HR ]5 m– 15 m]
Medium Resolution ]15 – 100 m]
Moderate Resolution ]100 – 300 m]
ALOS/PRISM
ALOS/AVNIR-2
QB
ASTER
MERIS MODIS
IKONOS
SPOT5 HRS/HRG
Landsat TM, ETM+, SPOT-HRV, IRS, CBERS
CARTOSAT-1/2 Worldview-1 Mission Sensor
Geoeye-1
SPOT 4 Kompsat-1
Kompsat-2 Eros A1 Orbview-3 Formosat-2 Eros-B
VEGA Technologies SAS © 2008 VEGA Group PLC
Page 13 of 54
CALIB-TN-WP222-GAEL_ETH Issue 1.0
Calibration Test Sites Selection and Characterization Methods for Calibration/Validation and Test site requirements
3.
DEFINITION
3.1
Fundamental concepts
3.1.1 Calibration vs. Validation The calibration / validation activities are primarily set with regard to the parameters to be estimated. A calibration / validation activity aims at following a calibration and / or validation methodology. The methodology is applied and based on one or several methods. The methodology is based on a refined model (direct/indirect) and used as input a certain level of information from the product itself. The calibration process aims at determination of the sensor model parameters precisely. The validation process aims to check the quality of the data. According to the accuracy results obtained in the validation process, the calibration procedure might be repeated. On the other hand, the validation process can be applied for the methods as well. The various approaches are finally intended to calibrate the instrument / platform and to validate the value added products.
3.1.2 Georeferencing approaches Georeferencing is a process which provides the position and rotation information of an object or an event at a certain time in an earth reference frame as output. The concept of the georeferencing in photogrammetric applications provides the position and the rotation of the sensor at the time of imaging. The procedure is called also as sensor orientation. According to the availability and use of a priori knowledge and external data on the sensor parameters, there are three methods to obtain the EO parameters of an image:
the direct georeferencing,
the indirect georeferencing,
the integrated georeferencing.
With the direct georeferencing method, the EO parameters are obtained from external instruments on board of a sensor platform, such as GPS, inertial measurement units (IMU), star tracker, etc. The indirect georeferencing is the conventional way of obtaining the sensor orientation parameters. The parameters are computed in a mathematical solution using a number of GCPs. The indirect georeferencing is a post-processing method, while the direct georeferencing method can be used for online photogrammetric applications. A third approach to solve the georeferencing problem is called integrated sensor orientation. It is a combined solution which employs both the direct and indirect georeferencing methods. The EO parameters provided by the external measurement devices are used as input and improved in this post-processing method. The input parameters are stochastically weighted in the process (RD28).
Page 14 of 54
VEGA Technologies SAS
Calibration Test Sites Selection and Characterization Methods for Calibration/Validation and Test site requirements
CALIB-TN-WP222-GAEL_ETH Issue 1.0
In the context of calibration and validation of satellite optical imagery, the direct georeferencing is used for validation of the geometric accuracy of the images. A spatial forward intersection procedure is applied for estimation of the points’ object space coordinates. The indirect georeferencing method is usually used when sensor a priori information is not available. The integrated sensor orientation method is used for both the calibration and the validation purposes with the a priori knowledge of the sensor geometry.
3.1.3 Rigorous vs. generic sensor models Another categorization of the georeferencing methods can be done according to the basic foundations of the mathematical model, which is used to describe the geometric relationship of an image and the corresponding ground point. Two broad categories are described: rigorous sensor models and the generic ones. The rigorous sensor model reflects the sensor geometry according to the physical reality of the sensor. A priori knowledge on the sensor model parameters are required to define the relationship between the image and the object spaces. The parameters of a rigorous sensor model are grouped as interior and exterior orientation parameters. For geometrical calibration of satellite sensors, a rigorous sensor model should be used. There is no unique formulation of the rigorous models. The functional model should be defined according to the physical characteristics of the sensor. The basic formulation is based on the collinearity equation, which is well known in photogrammetry. A rigorous sensor model can be used to validate the sensor model parameters as well. In principle, a rigorous model can be used for direct, indirect, and integrated sensor orientation. However, in case of high-resolution satellite imagery, where Linear Array sensors are used, the problem of indirect georeferencing is very complex and often it is not possible to get a solution. Therefore, the rigorous sensor model can be applied only when sensor interior and exterior orientation parameters are available at least partially, even though the quality of the measurements does not represent the state-of-the-art. The generic sensor models are independent from a priori knowledge of the physical sensor conditions. The geometric metadata information is not necessarily used with the generic models. The generic models are in most cases only approximations and generally do not produce as accurate results as the physical models. The main advantage of using the generic models is that a sophisticated knowledge of the sensor geometry is not required and that they can be used in a fairly simple way. Some well-known examples of the generic models are 2D/3D polynomial functions, the Rational Function Models, 2D/3D affine transformation, and the Direct Linear Transformation model (RD28). The generic sensor models can be used for geometric validation purposes only. They are especially useful for validation of geometrically pre-processed images. They must be preferred when a priori knowledge or data on the physical sensor model is not available. The generic sensor model parameters can be estimated with the indirect georeferencing approach using ground control points. An alternative way of estimation is to extract the generic model parameters from a rigorous sensor model. Nowadays, this approach very often used by satellite image vendors, for the computations of rational polynomial coefficients (RPCs) of the Rational Function Model. The following steps are applied for the computation:
VEGA Technologies SAS © 2008 VEGA Group PLC
Page 15 of 54
CALIB-TN-WP222-GAEL_ETH Issue 1.0
Calibration Test Sites Selection and Characterization Methods for Calibration/Validation and Test site requirements
Generate a number of virtual GCPs distributed on a regular grid using direct georeferencing approach with the rigorous model
Compute the generic model parameters using the virtual GCPs in an indirect georeferencing approach
Table 3: Categorization of the satellite optical sensor geometries Rigorous vs. generic
Cal/Val Activity
Georeferencing method
Rigorous sensor model
Cal & Val
Direct, indirect, and integrated sensor orientation*
Generic sensor model
Validation
Indirect sensor orientation
* indirect sensor orientation is not possible in practice for the Linear Array sensor images
3.1.4 Rigorous sensor model parameters The rigorous sensor model parameters are generally analyzed in two broad categories;
the interior orientation (IO) parameters,
the exterior orientation (EO) parameters.
Both parameter groups are used in the mathematical definition of the relation between an image point and the corresponding object point on the ground. This relation is described by the collinearity equation, which is the key component of a physical sensor model. The physical sensor model aims at reconstructing the image rays in the object space. The collinearity equation assumes that a point on the ground, projection centre of the image, and the same point in the image are located on a line. The IO reconstructs the image rays with respect to the projection centre. The IO parameters define the interior geometry of the sensors and they are used to transform the pixel coordinates of image points into image coordinate system. The EO defines the position and rotation of an image at the time of imaging. It includes the position of the projection centre in the object coordinate system and the image rotation parameters. The EO parameters can be obtained in a photogrammetric space resection process (i.e. indirect georeferencing) using ground control points (GCPs) or measured by EO measurement devices (i.e. GPS, INS, star tracker). In the latter case, the systematic errors of the EO measurement devices and the relative alignments of these devices with respect to the imaging sensor are subject to the calibration procedure.
Page 16 of 54
VEGA Technologies SAS
Calibration Test Sites Selection and Characterization Methods for Calibration/Validation and Test site requirements
CALIB-TN-WP222-GAEL_ETH Issue 1.0
z κ Z
PC (Xc,Yc,Zc) P′ (x,y)
ϕ
y ω
x
c
Focal plane PP (x0,y0,-c)
Y
P (X,Y,Z)
X
Figure 1: Transformation from camera system to the object (ground) system. PC: projection centre, PP: principal point, c: camera principal distance (focal length), (x,y,z): axes of image coordinate system, (X,Y,Z): axes of object space coordinate system
3.2
Sensors’ geometric characterizations
The calibration parameters, which constitute the functional model of the rigorous sensor model, are defined according to the sensor’s geometrical characterization. The sensor geometries of the satellite optical sensors can be categorized as in Table 4. Table 4: Categorization of the satellite optical sensor geometries Characterization Optical System
Sensor interior geometry
Options
Notes
Single optical camera
e.g. IKONOS
Multiple optical cameras
e.g. ALOS/PRISM
Point-based sensors
Whiskbroom systems
Linear Array sensors
Single viewing angle pushbroom sensors
Frame Array sensors
Central projection
per
camera,
- Low-resolution Resolution
Spatial
- Medium-resolution - High-resolution - Very high resolution
Radiometric
VEGA Technologies SAS © 2008 VEGA Group PLC
Varies in the range of 8-16 bits
Page 17 of 54
CALIB-TN-WP222-GAEL_ETH Issue 1.0
Characterization
Calibration Test Sites Selection and Characterization Methods for Calibration/Validation and Test site requirements
Options
Notes
Spectral
Varies in the range of 0.45-0.95 µm for the satellite optical sensors. Spectral sensitivity of each CCD sensor is named according to the bandwidth (e.g. RGB, PAN, NIR, etc.)
Temporal
Defined with the revisit time
Sensor and platform synchronization
- Synchronous
Operation principles Stereo acquisition
- Asynchronous - Along-track - Across-track
In addition, the parameters given in Table 5 should be analyzed for the Linear Array CCD sensors. Table 5: Line geometries of the satellite Linear Array CCD sensors Characterization Line constitution Staggered arrays
TDI technology
Options - Single CCD chip forms an image line - Multiple CCD chips form an image line optional - TDI is used (the number of image lines defines the TDI level) - TDI is not used
The satellite digital optical sensors are presented mainly in three different formats: (i) Point-based sensors, (ii) Linear Array CCD sensors, and (iii) Frame (Matrix/Area) Array sensors. A point sensor images only a single point at any instant of time (RD19). The point-based electromechanical sensors acquire images in whiskbroom mode. They use rotating mirrors to scan the terrain surface from side to side perpendicular to the direction of the sensor platform movement, like a whiskbroom (RD20). The width of the sweep is referred to as the sensor swath. Advantages of whiskbroom scanners over other types of sensors are that they have simple overall design, wide field-of-view, and easier to calibrate due to small number of detectors. On the other hand, they have more moving parts, post-processing is required due to spatial incongruence, and they have more constraints in spectral and spatial resolution due to low integration time (RD21). Well known examples of satellite whiskbroom imagers are MSS on LANDSAT 1-5, TM on LANDSAT 4-5, ETM+ on LANDSAT 6-7, AVHRR on POES, SeaWiFS on SeaStar, and the GOES satellites (RD20). The Linear Array sensors use CCD detectors located along a straight line in the focal plane. There are several configurations of the arrangement of the CCD lines in the focal plane of a lens/optics. The Linear Array CCD sensors operate with the pushbroom principle. The sensor is located in the focal plane perpendicular to the platform’s motion. The perspective projection is applicable only in the across-track direction. With the pushbroom principle, one image line is acquired at an instant of time and stored one after the other to form a strip during the platform movement. All high-
Page 18 of 54
VEGA Technologies SAS
Calibration Test Sites Selection and Characterization Methods for Calibration/Validation and Test site requirements
CALIB-TN-WP222-GAEL_ETH Issue 1.0
resolution satellite optical sensors and the aerial Linear Array CCD sensors operate with the pushbroom principle. Most of the high-resolution earth observation satellite sensors in operation (e.g. SPOT 4&5 sensors of CNES, France; IKONOS and Orbview-3 sensors of GeoEye. U.S.A.; KOMPSAT-1 and KOMPSAT-2 sensors of KARI, South Korea; QuickBird and Worldview-1 sensors of Digital Globe, U.S.A.; EROS-A1 and EROS-B sensors of ImageSat Intl., Israel; Cartosat-1 and Cartosat-2 sensors of ISRO, India; and PRISM and AVNIR-2 sensors of JAXA, Japan) and planned for near future (e.g. RapidEye sensor of RapidEye AG, Germany; GeoEye-1 sensor of GeoEye, U.S.A.; Worldview-2 sensor of Digital Globe, U.S.A.; EROS-C sensor of ImageSat Intl., Israel; and Pleiades-1 and Pleiades-2 sensors of CNES, France) are using Linear Array CCD technology. In digital frame sensors, the CCD pixels are positioned in a rectangular matrix. Similar to the film cameras, the images are taken in a central projection. The perspective projection is valid in all directions of imaging. In satellite sensors, the matrix array configuration is mostly applied in small-satellite missions (body mass < 1000 kg), e.g. Bilsat-1, TUBSAT series, UoSAT series, Kitsat series, etc., and in some of the meteorology and environmental monitoring satellite sensors, such as MERIS on ENVISAT, POLDER on ADEOS, etc. In comparison to satellite Linear Array CCD sensors, frame array CCD cameras of small satellite missions have larger GSDs or smaller coverage area due to small number CCD detectors (e.g. 2048 × 2048 pixels in Bilsat-1, 750 x 580 pixels in DLRTUBSAT, 1024 x 1024 pixels in UoSAT SHI and MSI cameras).
3.3
List of the Cal/Val parameters
The parameters defining the image geometry are calibrated in order to ensure that the geometric accuracy of the product and the stability of the sensor meet the requirements. The rigorous sensor model parameters are subject to the calibration activity. The IO parameters of the rigorous model are identified per sensor. The most common parameters are listed as in theTable 6. The IO parameters are initially determined in a laboratory environment using vicarious methods. The detailed definitions of the EO parameters of the rigorous model are also identified according to the sensor exterior geometry. For the high resolution satellite sensors, these parameters are measured with external measurement devices, such as GPS and star tracker. A main categorization is provided in Table 6. A third major category to be considered for the calibration of satellite sensors is the systems’ overall geometric accuracy. This term is mainly related with the quality of the image data and value-added products and subject to the validation activity. Table 6: List of parameters to be estimated Parameter to be Estimated
Interior orientation parameters
VEGA Technologies SAS © 2008 VEGA Group PLC
CALibration /VALidation
Reference
Principal point location
Cal
3.3.1.1
Camera constant
Cal
3.3.1.3
CCD Line position/rotation
Cal
3.3.1.3
Pixel size
Cal
3.3.1.3
Lens distortions
Cal
3.3.1.2
Page 19 of 54
CALIB-TN-WP222-GAEL_ETH Issue 1.0
Calibration Test Sites Selection and Characterization Methods for Calibration/Validation and Test site requirements
Parameter to be Estimated
Exterior orientation parameters
Reference
Attitude data
Cal/Val
3.3.2.1
Ephemeris (satellite position) data
Cal/Val
3.3.2.2
Relative alignments of the sensors
Cal
3.3.2.3
Direct geolocation
Val
3.3.3.1
Cal/Val
3.3.3.2
Local coherence
Val
3.3.3.3
Stereoscopic capability
Val
3.3.3.4
DSM product testing
Val
3.3.3.5
Ortho product testing
Val
3.3.3.6
Value added product testing
Val
3.3.3.7
Band-to-band registration Overall geometric accuracy
CALibration /VALidation
The descriptions of the parameters are given in detail in the following sections. Please note that the most of these assessments, except the Value added product testing, are performed on raw imagery in order to control the error.
3.3.1 Interior orientation parameters The interior orientation parameters of a satellite optical sensor can be categorized in two main groups:
Optical system related parameters
Imaging sensor (CCD detectors) related parameters
The number of optical systems used in a satellite sensor is important for the calibration task. If there are multiple optical systems, such as the 3 cameras of ALOS/PRISM sensor (forward, nadir, backward), each camera system should be calibrated individually. The imaging parts of a camera should be calibrated with consideration of the number of detectors, detector size, and the positions of each detector in the image plane. The IO reconstructs the image rays with respect to the projection centre. The IO parameters define the interior geometry of the sensors and they are used to transform the pixel coordinates of image points into image coordinate system. Although the IO parameters are identified per sensor, the most common parameters can be listed as in theTable 6. The IO parameters are usually determined in a laboratory environment using vicarious methods.
3.3.1.1
Optical system parameters - Principal point and distance
The principal point is defined as “that point on the image plane which is at the base of the perpendicular from the ‘centre of the lens’, or more correctly, from the rear nodal point” (RD16). The length of that perpendicular is the principal distance and, at infinity focus, it is equal to the focal length of the lens (RD17). The principal point coordinates and the principal distance (camera constant) are used in defining the image coordinate system, as depicted in Figure 2.
Page 20 of 54
VEGA Technologies SAS
Calibration Test Sites Selection and Characterization Methods for Calibration/Validation and Test site requirements
CALIB-TN-WP222-GAEL_ETH Issue 1.0
Precise calibration of these parameters is important for accurate georeferencing of the images. Geometric definitions of the principal point, camera constant (c) are shown in Figure 2.
Optical principal planes H H′ Object space
Optical axis τ Object point
Image plane
Image space
N=O N′=O′
τ′=τ
Central ray s Object distance
e
Image point
PP Principal point
s′=c Image distance
Figure 2: Idealized geometric image formation of an optical system. N, N′′ = Nodal points = projection centres O, O′′ (RD15).
3.3.1.2
Optical system parameters – Lens Distortion Parameters
The widely used definitions of the lens systematic distortions include the following terms
radial symmetric,
lens distortion,
decentreing distortion parameters
The Brown’s model (RD18]) is used largely in photogrammetric applications to correct the systematic errors in the image coordinates caused by the lens distortion.
3.3.1.3
Imaging sensor parameters
The imaging sensor parameters should be defined according to the sensor geometry. The main requirement of the calibration process is to determine the image space coordinates (x,y) of each CCD detector. The imaging sensor geometry can be defined with indirect parameters, which allows the determination of CCD detector positions with appropriate transformation functions. The most common parameters can be listed as:
Pixel size: dimension of one CCD detector is usually expressed in microns or millimetres
Number of CCD detectors (row and column)
CCD line inclination (angle)
VEGA Technologies SAS © 2008 VEGA Group PLC
Page 21 of 54
CALIB-TN-WP222-GAEL_ETH Issue 1.0
Calibration Test Sites Selection and Characterization Methods for Calibration/Validation and Test site requirements
CCD sensor size: length of the Linear Array CCD chip or height and width of the Frame Array CCD in millimetres.
The imaging sensor parameters should be defined according to the sensor geometry. The figures hereafter depict some typical sensor geometry. Figure 3: Focal plane arrangement of the IKONOS sensor (RD36)
TDI
Each channel consists of 3 CCD parts forming a virtual line, the middle part is shifted
Image of the focal plan
Figure 4: Quickbird and SPOT HRG sensors, focal plane arrangement (RD36)
Arrangement of CCDlines in focal plane of QuickBird above: panchromatic, below: multispectral (RD38)
Interior geometry of the HRG sensors. Both HRG sensors are identical. The images of the two lines (HMA and HMB) are resampled to produce 2.5 resolution PAN images, called THR (Très Haute Résolution or Very HighResolution).
3.3.2 External orientation parameters The EO defines the position and rotation of an image at the time of imaging. It includes the position of the projection centre in the object coordinate system and the image rotation parameters. The EO parameters consist of position of the camera projection centre (Xc,Yc,Zc) and the rotation angles of the image (ω,ϕ,κ) in the reference object space coordinate system at the time of the imaging (Figure 1).
Page 22 of 54
VEGA Technologies SAS
Calibration Test Sites Selection and Characterization Methods for Calibration/Validation and Test site requirements
CALIB-TN-WP222-GAEL_ETH Issue 1.0
The EO parameters can be calculated from the state vectors, which consist of position, attitude, and velocity vectors of the satellite contained in the product format. They are used to calculate the orientation of the orbital system with respect to the ECI and ECR systems.
3.3.2.1
Positional accuracy and stability The position data of the satellite platform is recorded by the GPS (Global Positioning System) receivers. A broad overview of GPS errors is provided in Table 7. The largest error is due to the receiver clock. Table 7: Error sources in GPS positioning (RD23) Error Source
Typical Magnitude
Receiver clock error (synchronized)
1 µs (300 m)
Residual satellite clock error
20 ns (6m)
Satellite synchronization to UTC
100 ns (30 m)
Selective Availability (cancelled by 2001)
100 m
Orbit error (precise, IGS)
20 cm
Tropospheric delay
20 (dense GCP test site)
Stereopair or triplet
9 per image
1 AP/image
5
>20 (dense GCP test site)
Stereopair or triplet
18 per image
1 AP/image
10
>30 (dense GCP test site)
Number of images
Recommended number of GCPs for comprehensive statistical analysis
* the computations are performed with the assumptions of:
A rigorous sensor model with different numbers of trajectory modelling parameters and with self-calibration (varying numbers of APs)
One image stereopair or triplet is used
Poor trajectory accuracy
Well distributed and highly accurate full GCPs (the planimetric and height coordinates are available with high accuracy)
No additional image points (i.e. tie points) are used
The trajectory parameters given in the table above includes the exterior orientation unknowns, which are the image position and rotation elements. These parameters can be modelled with a complex trajectory model, such as piecewise polynomials model. The number of trajectory correction parameters should be computed
VEGA Technologies SAS © 2008 VEGA Group PLC
Page 45 of 54
CALIB-TN-WP222-GAEL_ETH Issue 1.0
Calibration Test Sites Selection and Characterization Methods for Calibration/Validation and Test site requirements
according to the complexity level, e.g. number of trajectory segments, polynomial order, etc. On the other hand, the self-calibration parameters contain the additional parameters, which reflect the interior orientation parameters to be calibrated.
4.4.2 Validation In the frame of validation / verification the geometric model can be defined using a generic or generalized model. The rigorous model is not mandatory. The following table listed the number of GCPs required according the model to be estimated. Table 9: The number of unknowns in the widely used generic models and the corresponding minimum number of GCPs required for the solution (RD24). Model
Number of unknowns
Minimum number of GCPs
2D Affine transformation
6
3
3D Affine transformation
8
4
DLT model
11
6
Rational Functional Model (RFM)
80
40
2nd order 2D polynomial function
12
6
2nd order 3D polynomial function
20
10
3rd order 2D polynomial function
20
10
3rd order 3D polynomial function (RFM)
40
20
The distribution of GCPs is again key aspect; RFMs are sensitive to the distribution of the ground control points. A validation exercise relies on the use of the physical model to generate a set of tie points. These tie points will be corrected thanks to reference equipment (simple shift or more complex affinity transformation). The tie points will then be used to estimate a direct sensor orientation based on the RFM generic model. The comparison between the accuracy provides with the corrected generic model and the physical model is then performed. It may highlight some effects the physical model account whereas the generic model does not. The effects are mainly due to the attitude variation because of their high frequency [RD4].
4.5
Summary The number of model unknowns is not the same for the different sensors (point, linear array, multiple linear arrays). Accounting that for the orientation of one CCD chip image a minimum of four GCPs located at the every corner of the image are required, the number of GCPs depends on the number of the CCD chips. The linear arrays get more stable internal orientation parameters and are less challenging to calibrate. The effort is mainly focused on the external orientation parameters (alignments, attitude model). The more accurate the sensor spatial resolution is, the more critical is the calibration of the focal plan. For the high resolution sensor class (10 m GSD), the estimate of the internal orientation parameters or focal plan calibration is not challenging.
Page 46 of 54
VEGA Technologies SAS
Calibration Test Sites Selection and Characterization Methods for Calibration/Validation and Test site requirements
CALIB-TN-WP222-GAEL_ETH Issue 1.0
For very high resolution sensor, the focal plan calibration required a super site. A focal plan model can limit significantly the number of GCPs required for calibration [RD45]. However, the method needs very accurate GCPs that are ‘rare and precious’. So, CNES propose to use in first a field calibration method to calibrate focal plan and output. The focal plan calibration based on few GCPs will be more likely in operation for location model performance monitoring. The in flight correction aims at analyzing of the residual with regards to the model) will be performed. This approach based on focal model can be applied to the high/medium/low resolution sensor class. The mathematical model required number of GCPs that depends on the availability of scene orientation information; completely, partially or even not. The direct orientation method is used for the assessment related to the estimate of the direct location accuracy. For an agency mission, the scene orientation information is generally well-known. Thanks to the pre flight calibration (focal plan/camera parameter) and on board devices that provides accurate ephemeris and attitude measurements (stat tracker, gyro, three axis stabilization techniques), the image orientation process can be performed with indirect georeferencing method but with strong hypothesis on the accuracy of the auxiliary data. It limits significantly the number of GCPs to be used. The document demonstrates that the calibration method set up define the test site configuration. A first approach is empiric and historically used for calibration, this is the field calibration. This methodology is very constraint regarding the number of GCPs to be used and the distribution of GCPs. A second approach is statistical (stochastic model), this is the self calibration. The constraint on the number and distribution of GCPs is less. The constraints on the test field properties are then less important. For the both methodologies, the vertical distribution is also a key aspect, if the working scene is planned the geo referencing can be done but the calibration of all parameters not. For the major part of sensor class, the efforts shall be focused on
the estimate and monitoring of EO parameters, the estimate of the alignment errors, the band to band or image to image calibration.
Due the improvements of the technology, the ephemeris and attitude measurement is now done frequently. To check ephemeris, GCPs should be distributed in along track. To check the attitude the GCPs should be distributed anyway in the image, not necessary in the middle of the image, more in across track direction direction. A minimum of four GCPs (the four corners) is required to achieve the planimetric accuracy. It is better that the spacing in line direction between two neighbored GCPs exceed the time required between two attitude measurements. With only four GCPs, the systematic planimetric error of the model can be estimated but without control on the error model. Considering IO parameters stable, the difference between the EO parameters estimated through the model and those known a priori (auxiliary data) leads to correction of alignment errors; systematic bias (ephemeris data) and bore-sight alignment (attitude data). These corrections are finally global and do not indentify which part of the model failed. It has not been discussed into this report. The image to image calibration is a key issue as well. The band to band calibration it does not required accurate reference data. There is no equipment constraint associated to this validation item. The image can only rectified, no ortho rectification process. Within this hypothesis a flat terrain shall be selected. Moreover, the site properties shall be selected carefully; method is based on image correlation technique and so highly depends on the sensor class. The level of information to be matched is spatial resolution dependant… All these remarks shall be mitigated in VEGA Technologies SAS © 2008 VEGA Group PLC
Page 47 of 54
CALIB-TN-WP222-GAEL_ETH Issue 1.0
Calibration Test Sites Selection and Characterization Methods for Calibration/Validation and Test site requirements
case of hyperspectral instrument; the number of bands is important, and force scientist to consider a parallax effect that shall be compensated using an accurate digital elevation model. The image matching technique is used, as well, for the calibration/validation of the alignment between CCD chip (MERIS), radiometers (PRISM), or several instruments on the same platform (SPOT, IRPS6). In this case, the ortho processing shall be performed and an accurate equipment and terrain model is better. The payload steering mirror calibration is calibrated using image matching technique as well (IRPS6, AVNIR-2), it is more related to the calibration of EO/IO parameters. Two categories of image matching sites (interband specific and the ‘other’ ones) have been highlighted. The properties of the sites are different. The DSM testing site is obviously not a band to band calibration site but can be used to assess the other calibration parameters. However, some parameters are time varying, several sites are required and it is not cost effective to set up several DSM testing sites. The table just here after is a first attempt to propose the list of the parameters to be estimated against the methodology requirements and site selection. Table 10: List of the parameters Parameter to be Estimated
Calibration GCP distribution /validation
Georeferencing
Working data set selection
Calibration Homogeneous
In direct
Dense, 1 super site
CCD Line Calibration Homogeneous position/rotati on
In direct
Dense, 1 super site
Pixel size
Calibration Homogeneous
In direct
Dense, 1 super site
Lens distortions
Calibration Homogeneous
In direct
Dense, 1 super site
Ephemeris system
Calibration Homogeneous
In direct/Direct
Intermediate Along track
Calibration Across Track GCP located in Along Track for indepth analysis
In direct
Intermediate Along track
Relative alignments
Calibration Homogeneous
In direct
Intermediate Along track
Direct geolocation
Validation
Direct
Intermediate Along track
Camera constant
Interior orientation parameters
Exterior orientation Attitude parameters system
Overall geometric
Page 48 of 54
Additional GCPs located in Along Track for in-depth analysis
Homegeneous
VEGA Technologies SAS
Calibration Test Sites Selection and Characterization Methods for Calibration/Validation and Test site requirements
Parameter to be Estimated
CALIB-TN-WP222-GAEL_ETH Issue 1.0
Calibration GCP distribution /validation
Georeferencing
Working data set selection
Validation
No need in GCP
Direct
No reference data.
Band to band Calibration No need in GCP registration Relative Method Reference DEM
Direct
Image matching
accuracy Local coherence
VEGA Technologies SAS © 2008 VEGA Group PLC
Page 49 of 54
CALIB-TN-WP222-GAEL_ETH Issue 1.0
Calibration Test Sites Selection and Characterization Methods for Calibration/Validation and Test site requirements
APPENDIX 1: LIST OF THE EXISTING PROGRAM The table just hereafter lists the calibration programs highlighted in the work package 210. Table 11: Calibration program Site Name
Number of GCP collection GCPs method
Centre geo coordinates lat (dd) / lon (dd)
Phoenix – Arizona (US)
113
Stennis Space Centre (US)
46
Brookings – Dakota (US)
San Diego Seven (US)
-DGPS
Altitude Range (m)
33°26.7’ N
Calibration Parameters
Site Coverage Size NS X EW
EO
112° 03.9’ W
96
44.185 N
EO
Large Scale aerial photography
32.429 N
??
-Large Scale aerial photography (PELICAN) – GSD Resolution 50 cm
43.499 N
Manosque (France)
QUICKBIRD
VHR
IKONOS
VHR
[RD2] [480-500]
EO
IKONOS, QUICKBIRD, ORBVIEW [RD8]
VHR
[0-100]
EO
20 km x 30 km
IKONOS, QUICKBIRD ORBVIEW [RD2]
VHR
[200-600]
IO
30 km x 60 km
SPOT5 HRS/HRG, HR [RD3]
[500-1250]
EO/IO
3 zones: ALOS/PRISM -Bern: 110 km2 IKONOS -Thun: 100 km2 -Southwest: 90 km2 SPOT-5 [RD5]
96.468 W 140 Cultural features
Sensor class
[RD7]
-Circular Target (2.44 30°23 N diameter) 89° 37.6’ W -DGPS
Cal/val Missions
117.122 W
5.479 E
-DSM – IGN BDZ Resolution 1m 138 Bern/Thun (Switzerland)
-DGPS -DSM (from Aerial Photo)
46.839 N 7.533 E
DSM Testing Ortho Testing
Page 50 of 54
VHR HR
VEGA Technologies SAS © 2008 VEGA Group PLC
Calibration Test Sites Selection and Characterization Methods for Calibration/Validation and Test site requirements
Site Name
Number of GCP collection GCPs method
CALIB-TN-WP222-GAEL_ETH Issue 1.0
Centre geo coordinates lat (dd) / lon (dd)
Zurich/Winterthur
99
DGPS
47.531 N
Dense site
DSM generation from 8.684 E aerial photos
Altitude Range (m)
Calibration Parameters
Site Coverage Size NS X EW
Cal/val Missions
Sensor class
400-850 m
EO/IO
GCP: 35 × 35 km2
ALOS/PRISM
VHR
DSMTesting
DSM-1: 22.6 × 11.4 km2
[RD24]
Ortho Testing
(Switzerland)
Wellington (South Africa)
DSM-2: 14.6 ×9.4 km2 67
-DGPS
(Along Path Site) 67
33.508 S
[50-400]
EO/IO
35 km x 35 km
Adana
DGPS
31.531 N
0-1150 m
EO/IO DSMTesting
2 sets of DSM Lidar DSMs with 5 m 130.765 W data and 10 m grid spacing 75
-DGPS
36.777 N
(Along Path Site)
35.310 E
~120 (west) -RTK GPS (west) and ~80 (east) -1:2 500 scale maps Dense Site (east)
36.1 N
Tochigi
~90
37.0 N
(Japan)
Dense Site
Iwate
~90
(Japan)
Dense Site
Tomakomai
~45
(Japan)
Dense Site
(Turkey) Tsukuba (Japan)
VEGA Technologies SAS © 2008 VEGA Group PLC
VHR
GCP: 30 × 30 km2
ALOS/PRISM
VHR
DSM: 3 × 3 km
Cartosat-1
18.914 E
Sakurajima (Japan)
ALOS/PRISM,
DGPS
Ortho Testing 0-100 m
EO/IO
35 km x 35 km
ALOS/PRISM
VHR
[0-1000]
EO/IO
60 km x 40 km (west) ALOS/PRISM, [RD1] 40 km x 40 km (east)
VHR
[0-2000]
EO/IO
25 km x 70 km
ALOS/PRISM, [RD1]
VHR
[0-2000]
EO/IO
25 km x 70 km
ALOS/PRISM, [RD1]
VHR
[0-1000]
EO/IO
10 km x 70 km
ALOS/PRISM, [RD1]
VHR
139.9 E
140.3 E DGPS
39.7 N 141.0 E
DGPS
42.7 N
2
142.0 E
Page 51 of 54
CALIB-TN-WP222-GAEL_ETH Issue 1.0
Site Name
Calibration Test Sites Selection and Characterization Methods for Calibration/Validation and Test site requirements
Number of GCP collection GCPs method
Centre geo coordinates lat (dd) / lon (dd)
203 Saitama Dense Site
-DGPS -DSM (Lidar)
35.918 N
Altitude Range (m)
Calibration Parameters
Site Coverage Size NS X EW
Cal/val Missions
Sensor class
[20-250]
EO/IO
40 km x 60 km
ALOS PRISM, [RD4]
VHR
20 km x 20 km
PRISM
VHR
KOMPSAT-2
VHR
AVNIR-2
HR
ALOS AVNIR-2
HR
139.490 E
DSMTesting
(Japan) Ortho Testing 35
-DGPS
46.839 N
-DEM (Spot)
7.533 W
[0-200]
La Crau
IO/EO/Direct location
(France) Le cap
35
(South Africa) World wide Image chip collection Tokyo
30 Scenes / 20-30 GCPs
20-30 GCPs
(Japan) Tokyo
Existing data – IKONOS
33.99 S
[23-70]
Direct location
70 km x 70 km
18.75 E N.A
[RD9] N.A
Direct location
70 km x 70 km
Topographic Map (1:2500 /1:5000)
N/A
N.A
N.A
Direct location
60 km x 60 km
ASTER
MR
[RD13] N.A
N.A
Interband
60 km x 60 km
ASTER
MR
[RD13]
Brookings – Dakota ? (US)
-Aerial photography (DOQQ)
44.185 N
Brookings – Dakota No (US)
-Aerial photography (DOQQ)
44.185 N
La Crau
Existing data – SPOT 443.5 N
Page 52 of 54
MR
[RD13]
Oman
(France)
ASTER
35
[480-500]
EO
185x185 k;
96.468 W
LANDSAT EMT+
MR
[RD6] [480-500]
Interband
185x185 k;
96.468 W
LANDSAT EMT+
MR
[RD6] [0-1000]
Direct location
185 km x 370 km
LANDSAT TM
MR
4.98 E
VEGA Technologies SAS © 2008 VEGA Group PLC
Calibration Test Sites Selection and Characterization Methods for Calibration/Validation and Test site requirements
Site Name
Number of GCP collection GCPs method
CALIB-TN-WP222-GAEL_ETH Issue 1.0
Centre geo coordinates lat (dd) / lon (dd)
Senegal
Tie points
About 200
Image chip collection World wide Image chip collection
VEGA Technologies SAS © 2008 VEGA Group PLC
14.5 N
Calibration Parameters
Site Coverage Size NS X EW
Cal/val Missions
Sensor class
[0-400]
IO
370 km x 555 km
MERIS
Low resolution
Landsat ETM+ image 14.49 W synthesis
(Africa) Europe
-Existing data,
Altitude Range (m)
-Existing data,
N.A
[RD11] N/A
Direct geo location N/A
Landsat ETM+ image synthesis About 420
-Existing data,
MERIS
Low resolution
[RD12] N.A
N/A
Direct geo location N/A
Landsat TM+
MODIS [RD14]
Page 53 of 54
Low resolution
CALIB-TN-WP222-GAEL_ETH Issue 1.0
Calibration Test Sites Selection and Characterization Methods for Calibration/Validation and Test site requirements
APPENDIX 2: THE GEOMETRIC CALIBRATION PROCESS
Page 54 of 54
VEGA Technologies SAS © 2008 VEGA Group PLC